This computer program makes pharmaceutical patents ai

image: its graphical summary illustrates how, by identifying and keeping track of the major disconnects essential to patent-protected syntheses, a retrosynthesis computer program can autonomously design synthetic pathways, ”navigating around” approaches previously published or patented.
see Following

Credit: Molga et al./Chem

The manufacturing routes of life-saving drugs and other pharmaceutical compounds are some of the best-protected trade secrets in the global industry. Building on recent work programming computers to identify synthetic pathways leading to pharmaceutically complex molecules, researchers in Poland and South Korea have unveiled computer-based methods to uniquely suggest synthetic strategies that circumvent protected aspects. by essential medicines patents. Their work appears on January 17 in the journal Chemistry.

“When we started this project, I was somewhat skeptical about the machine’s ability to find viable synthetic alternatives – after all, these are blockbuster drugs worth billions of dollars, and I was sure the respective companies had covered the patent space so densely that no loophole remained,” says lead author Bartosz Grzybowski, professor of chemistry at the National Institute of Science and Technology in Ulsan, South Korea, and the Polish Academy of Sciences and lead developer of organic synthesis software Chematica “It turns out the loopholes are there, and we can find new routes for retrosynthesis that bypass patents entirely.”

These pharmaceutical patents protect the company’s intellectual property while preventing competing companies from using certain key synthetic solutions – painstakingly developed through experience to maximize yield, increase purity and reduce costs – when trying to produce the desired compounds. To arrive at viable unpatented options, the researchers “froze” difficult parts of each target molecule, forcing the computer to substitute unconventional but chemically plausible approaches based on mechanistic rules. They tested their system on three notable commercial drugs with different chemical barriers: linezolid, an antibiotic of last resort; sitagliptin, an antidiabetic; and panobinostat, a treatment for multiple myeloma.

In each case, when allowed to operate unconstrained, the program recommended commercial syntheses. But when even a few atoms and bonds were designated as untouchable, he broke new ground by applying Chematica’s existing functions to come up with new plans that carefully avoided those already patented. “By algorithmically locating the key bonds on which the patents are based and propagating them in Chematica’s retrosynthetic trees, we can generate synthetic solutions from alternative but economical starting materials, achieving real practical impact,” says Grzybowski. .

Chematica’s patent avoidance capabilities could also change the way chemists approach intellectual property and patent law. For example, computer-aided searches could be used to narrow down many different parts of a target molecule, bundling radically different syntheses into a single airtight patent. According to Grzybowski, however, such a patent would not necessarily remain without a loophole forever, thanks to the likely future experimental discovery of new reactions advancing chemical knowledge through healthy competition.

Overall, the researchers hope their software will help pharmaceutical companies better protect their intellectual property and, at the same time, help accelerate organic chemistry research and development by providing synthetic routes that are different from standard approaches. “This work exemplifies the benefits of pushing chemists to think algorithmically and requiring computer scientists to grasp key chemical concepts, delivering chemical artificial intelligence results that matter beyond the confines of academia.” co-author Piotr Dittwald, researcher with a background in mathematics. and computing.


This work was supported by the Defense Advanced Research Projects Agency (DARPA) and the Institute for Basic Science (South Korea).

ChemistryMolga et al. : “Navigate patented pathways preserving specific motifs along computer-planned retrosynthetic pathways”

Chemistry (@Chem_CP) is the first physical science journal published by Cell Press. A sister journal to Cell, Chemistry, which is published monthly, is home to fundamental and insightful research and shows how fundamental studies in chemistry and its sub-disciplines can help find potential solutions to tomorrow’s global challenges. Visit: To receive Cell Press media alerts, contact [email protected]

Warning: AAAS and EurekAlert! are not responsible for the accuracy of press releases posted on EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.

Gordon K. Morehouse